utf8_utf32.cc

#include <cstdint>
#include <optional>
#include <string>
#include <vector>
#include <iostream>

using utf8 = std::u8string_view;
using utf8_iter = std::u8string_view::const_iterator&;

#define CONT_MASK 0b00111111

constexpr auto utf8_first_byte(uint8_t byte, uint32_t width) -> uint32_t {
    return (byte & (0x7F >> width));
}

constexpr auto utf8_acc_cont_byte(uint32_t ch, uint8_t byte) -> uint32_t {
    return (ch << 6) | (byte & CONT_MASK);
}

constexpr auto unwrap_or_0(utf8_iter iter, utf8_iter end) -> uint8_t {
    if (iter == end) {
        return 0;
    } else {
        return *iter++;
    }
}

auto next_code_point(utf8_iter iter, utf8_iter end) -> std::optional<uint32_t> {
    if (iter == end) {
        return std::nullopt;
    }

    auto x = *iter++;

    if (x < 128) {
        return x;
    }

    auto init = utf8_first_byte(x, 2);
    auto y = unwrap_or_0(iter, end);
    auto ch = utf8_acc_cont_byte(init, y);

    if (x >= 0xE0) {
        auto z = unwrap_or_0(iter, end);
        auto y_z = utf8_acc_cont_byte((y & CONT_MASK), z);
        ch = init << 12 | y_z;
        if (x >= 0xF0) {
            auto w = unwrap_or_0(iter, end);
            ch = (init & 7) << 18 | utf8_acc_cont_byte(y_z, w);
        }
    }

    return ch;
}

auto convert(const utf8& bytes) -> std::u32string {
    auto iter = bytes.begin();
    auto end = bytes.end();

    auto out = std::u32string{};

    while(true) {
        auto next = next_code_point(iter, end);

        if (!next.has_value()) {
            break;
        }

        out.push_back(next.value());
    }

    return out;
}

auto main() -> int {
    std::u8string_view utf8 = u8"가나다라";
    auto utf32 = convert(utf8);

    for (auto ch: utf32) {
        std::cout << static_cast<uint32_t>(ch) << std::endl;
    }
}